This application proposes a second five-year follow-up evaluation of a sample of 338 ethnically diverse, young adults (males=148) between the ages of 24-29 years old to examine the 'Deviance Prone' model of vulnerability in relation to predicting several alcohol use behaviors, including pathological involvement. This model posits that a significant aspect of the susceptibility for developing pathological alcohol involvement due to a parental history of alcoholism may result from deficient socialization and problem behavior. Although the Deviance Prone model was originally proposed as a heuristic model to explain the development of pathological alcohol involvement, many elements of the model have also been implicated as predictors of early onset of alcohol use and early heavy use of alcohol. The proposed study builds upon and extends our previous work on conduct problems and antisocial personality disorder with this and other samples. Information on key features of the model were initially collected when subjects were 14-19 years old and again at 19-24 years old. The study is unique in that the sample was ascertained from the community, is ethnically diverse, and has been well characterized using clinical, psychosocial, cognitive, and electrophysiological measures during the two initial assessments; parents were also directly interviewed. Using data from the three different time points, the present proposal will: 1) test the accuracy of the model for predicting different drinking behaviors at three different points in the subjects' lives: in mid-adolescence, in late adolescence, and in young adulthood; 2) examine the influence of age, ethnicity, gender, peer, and family relations, and coping ability as potential mediators/moderators of the Deviance Prone model's ability to predict alcohol use and alcohol problems at these three different time points; and 3) examine the accuracy of the Deviance Prone model for predicting alcohol use behaviors over time. Using structural equation modeling methods, the data collected at baseline and at Time 2 will be used to predict drinking behaviors, including pathological alcohol use, at Time 3. We will also continue our analyses using the baseline data to predict alcohol use behaviors at Time 2. The stability and efficiency of the Deviance Prone model for predicting early initiation of regular alcohol use, heavy regular use, and pathological alcohol involvement at these three different points in time will also be examined.